Caldasia
17(2); 205 - 210, 1993
AN EVALUATION OF THE USE OF THE
DRY-WEIGHT-RANK AND THE COMPARATIVE YIELD
BIOMASS ESTIMATION METHODS IN PARAMO
ECOSYSTEM RESEARCH
G.M. HOFSTEDE
J.L. WITTE
ROBERT
HENK
Department
Kruislaan318,
o,
Systematics, Evolution and Paleobiology,
1098SMAmsterdam,
TheNetherlands.
Hugo de Vries-Iaboratory,
University
o,
Amsterdam,
Resumen
Se evaluó la aplicabilidad
de una combinación de dos métodos para estimar la biomasa
general y la composición
botánica, en una vegetación natural paramuna en el Parque
Nacional Natural los Nevados (Cordillera Central de Colombia).
El primer método (ecomparative yield») determina la biomasa general, destruyendo parcialmente la vegetación de los
cuadrantes de muestreo y el segundo (<<dryweight rank») determina la composición botánica
con base en el peso seco, sin destruir la vegetación. Estos métodos, inicialmente desarrollados
para pajonales forrajeros en Australia, se adaptaron para ser utilizados en el ecosistema
paramuno. Como resultado se obtuvo una estimación de la biomasa aérea de 2864 9 peso
seco. m·2 (desviación stándard 48) en la cual, la gramínea Calamagrostis
effusa contribuyó
con el 70%.
Puede concluirse que el método de producción comparativa es útil para estimar la biomasa
en el ecosistema paramuno, siempre y cuando se utilicen las adaptaciones
mencionadas.
Por otra parte la estimación de la composición botánica con este método dio resultados más
satisfactorios que con el método «dry weight rank», el cual presenta demasiados problemas
debido a la complejidad de los pajonales del páramo.
Abstraet
The use of the combination of the semi-destructive
comparative yield method for overall
biomass estimation and the non- destructive dry-weight-rank
method for studying botanical
composition on a dry weight basis in an undisturbed páramo vegetation in the Los Nevados
national park (Colombian Central Cordillera) was evaluated. These methods, developed for
Australian production grasslands, were adapted for use in the páramo ecosystem.
The
average above ground biomass in the area was estimated as 2864 9 dryweight. m-2 (sd.48),
ofwhich the bunchgrass Calamagrostis effusa contributed with ca 70%.
When used with some adaptations, the comparative yield method seems suitable for biomass
estimations in the páramo ecosystem. The here presented estimation of botanical eomposition with this method gave better results than dry-weight-rank
method, which had too many
shortcomings for use in the complex páramo grassland ecosystem.
Introduction
The most widely used method to estimate
aboye ground biomass in grasslands is the
destructive sampling of plots and subsequent
extrapolation to establish the yield of vegetation quadrats (Cochran 1963; 'TMannetje,
1978). The optimum quadrat size is that size
that provides the smallest confidence interval
ofthe mean for a given labour cost. When the
vegetation structure becomes more complex,
the optimum quadrat size and cost will in-
crease if the same confidence interval is to be
maintained (Wiegert, 1962).
The vegetation structure of neotropical equatorial alpine grasslands (páramos) is one of
the most complex among the grasslands in the
world (Cleef, 1981). The dominant growth
forms of the páramos in the Colombian Central Cordillera are stemrosettes
(Espeletia
hartwegiana subsp centroandina),
constituting an emergent (up to 5 m) vegetation layer
205
Caldasia Vol. 17, 1993
aboye high (up to 1 m) growing bunchgrasses
(Calamagrostis spp., Festuca spp), and dwarf
shrubs (Pernettya spp., Bacharis spp.). An
additional
lower vegetation
layer may be
present with ground rosettes (Hypochaeris
spp.) sedges (Carex spp.) and short-growing
grass species (Calamagrostis
spp., Agrostis
spp.).
Almost bare spots are common
(Salamanca, 1991). Due to this spatial heterogeneity and complex structure, optimum quadrat size for destructive sampling will be high.
For accurate biomass measurements ofpáramo
vegetation,
a less destructive
and faster
method, which takes into account the spatial
heterogeneity,
may be favourable.
In the
present study we evaluate the usefulness of a
combination of two methods to estimate biomas s and composition of Australian production grasslands, adapted for and applied in a
relatively undisturbed páramo vegetation.
The comparative
method
yield/dry weight rank
In the semi-destructive
comparative
yield
method for estimating total aboye ground
grassland biomass (Haydock & Shaw, 1975),
a series of sampling units (standard quadrats)
is selected subjectively and marked to construct an ordinal scale over the total range of
dry matter yields within a study area, using
the following procedure:
First, a sampling
unit with the highest dry matter yield of the
area under investigation (class 5) and a sampling unit with the lowest yield (not zero,
class 1) are selected.
Subsequently, a sampling unit in between 1 and 5 (class 3) and
sampling units in between 3 and 5 (class 4)
and in between 1 and 3 (c1ass 2) are selected.
A large number of sampling units (scored
quadrats) is selected at random in the area, of
which dry matter yield is given a score on the
scale of the standard quadrats. Depending on
the relative difference between the scale units,
intermediate scores are allowed. Finally the
five standard quadrats are harvested and the
material oven dried.
With the biomass of
these five quadrats and the scores of the randomly selected quadrats total biomass can be
estimated (Haydock & Shaw, 1975).
206
Both the selection of the standard quadrats as
the assignment of scores is based on visual
parameters as density, height and woodiness
of the vegetation. An initial training period is
necessary to calibrate visual selection and
scoring with actual dry weight and to minimise differences in scoring by different investigators (testing of scores by harvesting and
weighing, Friedel, et al., 1988).
Botanical composition on a dry weight basis
can be estimated with the non-destructive
dry-weight-rank
method ('TMannetje & Haydock, 1963). In this method, a large number
of quadrats are selected at random in an area.
In each an estimate is made ofthe species with
highest rank on a dry weight basis, by means
of the mentioned visual parameters.
Similarly the species with the second and third
rank are identified. For each species the load
of first ranks is multiplied by 8.04, second
ranks by 2.41 and third ranks by 1.00 (sen su
'TMannetje & Haydock, 1963). Summing the
resulting products gives a species-score, which
is then expressed as a percentage of summed
species-score of all species. This percentage
represents the botanical composition. "TMannetje & Haydock (1963) have shown that the
mentioned multipliers can be applied to any
type of grassland.
The two methods, inc1uding the multipliers,
have been applied in grasslands all over the
world, and have been tested and improved by
different authors (Friedel et al., 1988; Kelly
& McNeill, 1980).
Adaptations to the paramo ecosystem
In this study the comparative
yield/dryweight-rank combination was adapted to estimate overall aboye ground biomass (inc1uding
litter, adult stemrosettes were excluded) and
botanical
composition
of an undisturbed
páramo vegetation (Caño de Agua Leche, Los
Nevados National Park, Colombian Central
Cordillera at 4100 m alt., 4 48'N, 75 24'W).
In an area of 3 ha with a relative homogeneous
distributed vegetation five standard quadrats
(l m2) were se1ected and marked.
Next the
whole area was divided in a systematic
Hofstede
transeet-line pattern. While walking along all
transeets, 50 quadrats were randomly seleeted
by throwing (blind-eyes) a hoop (diam.100
cm). By this proeedure all points in the are a
had the same ehange to be sampled.
Eaeh
quadrat was given a seore on the standard
quadrat-seale, with a preeision of 0.25 (1.00,
1.25, 1.50, 1.75, ... 5.00). Although 50 quadrats should suffiee to minimise eonfidenee
limits (Friedel et al., 1988), two independent
data-sets of 50 were analysed to obtain a
better idea about the stability of the results.
Average seores of the eomparati ve yield
method ofthe 2 data-sets were 3.28 and 3.18,
with standard deviations of 0.88 and 1.04
(range: 1-4.75), refleeting the heterogeneity
of the vegetation. The seores approximate to
a normal distribution (p<0.05).
After harvesting and drying (24 hr at 105 C),
the yield of the 5 standard quadrats was
weighed. Total yields (dead and living, inel.
litter) amounted to 234, 808, 1788, 4275 and
9462 g DW.m-2. After a logarithmie (In) transformation of these values, the funetion In Y =
4.68 + 0.91 * Qe deseribed the relation between standard quadrat class (Qc) and In yield
(In Y), (R2 = 0.99, p<O,OI). This funetion
eould be applied to the s ores of the seores
quadrats to ealculate average dry weight.m-2
of the area:
& Witte: Yield biomass
50
L (0.91 * Sei
+ 4.68)
i=1
In B=--------
(1)
50
where: B= Average aboye ground biomass of
are a (g DW.m-2)
Sei = Seore of seored quadrat ion range 1-5
The average aboye ground biomass of the
data-sets 1 and 2 was 2836 and 2891 gDW.m-2
respeetively (average of both sets: 2864 g
DW.m-2, sd=48).
The estimates of the biomass presented here
were eompared with the yield oftwo independently seleeted and harvested seored quadrats
(seores: 2.75 and 3.25). With the regression
equation their biomass was estimated as 1316
and 2074 g respeetively.
Aetually they supported 1228 and 1966 g, an over estimation of
7%.
Dry-weight-rank estimates were also made in
the previous seleeted seored quadrats. Ranks
were assigned to 5 species and to 4 groups
(dwarf shrubs, ground rosettes, sedges and
short grasses were not ranked on species level).
Calamagrostis effusa eontributed 70% to total biomass of the area with data-set 1 and
74% with data-set 2 (table 1). For the other
Table 1. Load of first, second and third ranks per species, species scores and botanical composition (% of dry
weight) of the overall biomass, as obtained by the dry-weight-rank
method. Results of 2 data-sets of 50 scored
quadrats.
Data set 1
Species/groups
Calamagrostis
C.
recta
effusa
Sedges
Short grass
Ground rosettes
Rosaceae
Lupinus microphyllus
Dwarf shrubs
Juvenile Espeletia
Total
score
* 3rd
* 1st
#
ranks
ranks
ranks
42
5
1
1
O
1
O
O
O
24
13
4
2
1
3
O
2
1
6
6
8
4
4
10
1
4
7
2nd
Data set 2
Specles
% dry
#
seo re
welgh t
ranks
401.5
77.5
25.7
16.9
6.4
25.3
1.0
8.8
9.4
572.5
70.1
13.5
4.5
2.9
1.1
4.4
0.2
1.5
1.6
100.0
1st
42
5
O
O
O
3
O
O
O
Specles
% dry
ranks
ranks
score
welght
33
O
5
O
2
2
2
5
1
8
1
8
2
10
7
4
6
4
425.2
41.2
20.1
2.0
14.8
35.9
8.8
18.1
6.4
74.3
7.2
3.5
0.4
#
2nd
#
3rd
572.5
2.6
6.3
1.5
3.2
1.1
100.0
207
Caldasia
Vol. 17, 1993
species differences
were considerable.
between the two data-sets
Another botanical composition
estimation
procedure was developed to be used with the
comparative yield method. After harvesting
the standard quadrats the yield was separated
in the different species/groups,
which were
then subdivided in living and dead material.
The contribution
of a particular unit (e.g.,
Calamagrostis effusa, young leafs), expressed
as a percentage of the total yield of a standard
quadrat was determined.
Subsequently, botanical composition was ca1culated as:
5
2,
(%Sc(a+x)*(%Vt(a)*(l-x)
+ %Vt(a+l)*x)}
a=I
%Vt -----------(2)
100
where % Vt = Contribution (%) of unit to the
overall biomass in the are a
%Sc(a+x) = Contribution (%) of score a+x to
total number of scored quadrats
a = Integer part of score (l,2,3,4,5)
x = Fractional part of score (if a 4, x = 0.25,
0.5,0.75; else x = O)
% Vt(a) = Contribution (%) of unit to the yield
of standard c1ass a.
%Vt(+1) = Contribution
(%) of unit to the
yield of standard quadrat c1ass a+ 1
This method resulted in a more detailed overall botanical composition than the one obtained by the dry-weight-rank method: to more
species a contribution was attained and a subdivision per unit was made in living and dead
standing crop (table 2).
Discusion and conclusions
Rigourous statistic testing was not possible
due to of the small sample size, the lack of
standard quadrat replicas and the application
of the methods in just one páramo site. Moreover, data on biomass from other regions are
not available so comparisons of the present
overall biomass figures with the results of
208
other methods can not be made.
less, some remarks can be made.
Neverthe-
The comparative yield method seems suitable
for páramo ecosystem research. Because of
the big difference between the yields of different spots within the vegetation it is easy
even for relative inexperienced observers to
select the 5 standard quadrats covering the
biomass range, and to score vegetation quadrats on this scale. This is suggested by the
small diference between the two sets and by
the accuracy of the estimated yield of the two
sampled scored quadrats. A in-transformation
of the yields allows the use of a linear regression model to describe the relation between
yields and scores, but one yield per c1ass is too
few for statistical analysis. More replicas per
class or more classes, possibly with a smaller
quadrat size, would improve the significance.
The variance of data-set scores was quite
high, which reflects the inherent heterogeneity of the vegetation and its biomass. Analysing the distribution of the scores with respect
to additional parameters (terrain factors, land
use) recorded during the scoring of the quadrats could probably provide useful information about the distribution of the biomass over
the area.
Although the composrtion
of the standard
quadrats was very diverse, the estimates of
overall botanical composition with the comparative yield method resulted in detailed figures.
Results of this kind, or even more
precise, can be obtained using conventional
methods, but that involves the destruction of
far more quadrats and a considerably larger
time investment.
The dry-weight-rank
method for estimating
botanical composition seems less appropriate
in this vegetation, with its complex structure,
especially as information on the living/dead
ratio s can not be obtained with this method.
Due to the dominance of Calamagrostis effusa in almost al! scored quadrats, the contri bution of the other species to overall biomass
was negligible. Of the 150 rank scores made
in each data-ser, 75 pertained to Calamagrostis effusa and 75 to all 8 other species/groups.
Hefstede & Witte: Yield biemass
Table 2. Botanical composition (% of dry weight) ofthe five standard quadrats and ofthe overall biomass (results
of 2 data- sets of 50 scored quadrats), as obtained by the comparative yield method.
Botanlcal composltlon or standard quadrats (%)
Overoll botan leal
composltlon (%)
Unit
Living
Calamagrostis effusa young lea ves
effusa old leaves **
effusa sheath
effusa shoots
recta young leaves *
recta old lea ves **
recta sheaths
C.
C.
C.
C.
C.
C.
Sedges
Short grass
Ground rosettes
Rosaceae
*
Standard
Standard
Standard
Standard
Quadrat 1
Quadrat 2
Quadrat 3
Quadrat 4 Quadrat 5
1.0
1.6
1.3
2.6
4.5
5.0
2.5
4.5
12.4
0.8
0.4
0.5
1.8
8.5
1.1
4.1
7.9
3.0
6.2
0.5
5.6
Lupinus microphyllus
2.2
4.7
0.5
23.2
11.5
11.7
2.7
1.1
0.2
2.8
0.1
Other herbs 1.00.50.30.3
Dwarf shrubs
32.4
Juvenile Espeletia
2.2
0.2
3.7
15.8
13.7
Standard
1.1
Sedges
Short grass
Juvenile Espeletia
17.4
14.1
10.6
12.8
1.0
1.7
2.6
1.0
2.3
3.8
8.1
0.1
0.1
0.2
0.7
2.8
0.1
1.7
1.3
4.1
0.3
0.2
2.2
3.5
7.9
0.1
0.1
0.2
0.5
0.1
2.2
1.2
1.1
0.3
5.7
0.3
n.o
12.1
4.8
0.4
4.7
0.3
3.8
0.2
0.1
2.9
0.1
0.1
0.1
Date-ser
2
0.8
Dead
Calamagrostis effusa leaves
C. effusa sheaths
C. recta leaves
Data-set
Litter
.32.7
10.7
41.8
3.6
42.7
14.7
54.4
Other
Total
100.0
100.0
100.0
100.0
Calamagrostis spp
* Overall
** Green
33.6
0.2
100.0
45.6
6.7
44.4
100.0
100.0
6.1
green leaves are defined as young leaves.
leaves with yellow spots are defined as old leaves.
This is reflected in a low difference of the
proportions of Calamagrostis effusa and high
differences of the proportions of the other
species between the two data-sets (Table 1).
Both the results of the botanical composition
obtained in the comparative
yield method
(summation ofliving and dead, litter exc1uded)
and the results obtained in the dry-weight-rank
method, show percentages of Calamagrostis
effusa in the same range (CY:67% -set 1- and
62% -set 2-; DWR 70 and 74% respectively).
All other species showed different contributions in the results of the two methods.
The comparative yield method, with the mentioned adaptations, could be a fast and reliable method of estimating
biomass
and
composition of complex grasslands, if tested
more intensively. The execution ofthe present
study was ca. 24 hours, inc1uding harvesting,
subdividing and weighing, which may be considered fast to obtain this kind of information
209
Caldasia Vol. 17. 1993
(Wiegert, 1962). When the sample set is
increased as mentioned (more classes or more
replicas per c1ass) nonparametric regression
and bootstrap analysis (Efron & Tibshirani,
1991) could be a recommendable test. An
important additional reason for using it in
threatened systems is that the methods requires less quadrats to be destructed than
conventional methods. The dry-weight-rank
method seems to have too many shortcomings
to be practical in the páramo ecosystem.
Acknowledgements
The authors would like to express their thanks to
the Instituto Nacional de los Recursos Naturales
Renovables y del Ambiente (Inderena; Bogotá
Manizales) for providing hospitality in the Los
Nevados National Park. Dr. Alejandro Velázquez
Montes and G. Guillot M. Se. are acknowledged
for their constructive
criticisms. Dr. Sonia
Salamanca improved the spanish abstract.
The first author was supported by grant W84-283
ofthe Netherlands Foundation for the Advancement
of Tropical Research (WOTRO).
Literature
cited
CLEEF,A.M. 1981. The vegetation of the páramos of the Colombian Cordillera Oriental.
210
Dissertaciones
botanicae 16.J. Cramer,
Vaduz.
COCHRAN
W.G. 1963. Sampling Techniques. John
Wiley & Sonso New York.
EFRON,B. & R. TIBsHIRANI.1991. Statistical analysis
in the computer age. Sclencie253:390-395.
FRIEDEL,M. H., V.H. CHEEWINGS
& G.N. BASTIN.
1988. The use of comparative yield and
dry-weight rank techniques for monitoring
grassland. J. Range Manage. 41: 430-435.
HAYDOCK,
K.P. & N. H. SHAW.1975. The comparative
yield method for estimating dry matter yield of
pastures. Aust. J. Exp. Agr. Anim. Husb. 15:
663-670.
KELLY, R.O. & L. McNEILL. 1980. Tests of two
methods for determining herbaceous yield
and botanical composition.
Proc. Grassl.
Soc.Afr.15: 167-171.
'TMANNETJE,
L. 1978. Measurement of grassland
vegetation and animal production. Commonw. Agricult. Bureaux. Farnham Royal,
Bucks.
___
• & K.P. HAYDOCK. 1963.
The dryweight-rank method for the botanical analysis
of pasture. J. Brit. Grassl. Soco 18: 268-275.
S. SALAMANCA
V. 1991. The vegetation of the
páramo and its dynamics in the volcanic
massif Ruiz-Tolima (Cordillera Central, Colombia) Ph. D. Dissertation, University of
Amsterdam.
WIEGERT,R.G. 1962. The selection of an optimum
quadrat size for sampling the standing crop
of grasses and forbs. Ecology43: 125-129.